World Congress 2024 • Aug 29, 2024 • Session details

Minimising the Carbon Footprint of Workloads

Michael Mueller

Idling servers still consume 60% of their peak power. Slash your infrastructure's carbon footprint using kernel-level energy profiling, strict right-sizing, and strategic workload time-shifting.

Pause
Mute Enter Fullscreen
#1 about 3 min

Rising technology impact on global overall carbon emissions

Global carbon dioxide emission goals directly intersect with the projected tripling of cloud computation output.

#2 about 3 min

Environmental cost of large artificial intelligence language models

Training and running automated inference for generative language models requires massive and unsustainable global power output.

#3 about 2 min

Navigating incoming infrastructure grid power and efficiency regulations

Emerging constraints on localized power grids alongside incoming energy efficiency laws completely alter future computing center builds.

#4 about 2 min

Carbon transparency reporting issues from centralized cloud providers

Late external reporting systems and a consistent lack of real-time analytics currently prevent application engineers from optimizing baseline carbon emissions.

#5 about 3 min

Optimizing active server utilization limits and power effectiveness

Pushing operational hardware utilization percentages closer to optimal thresholds balances performance capacity against heavy data center power requirements.

#6 about 2 min

Reducing embedded carbon footprints by extending hardware lifecycles

Delaying continuous physical hardware replacements actively minimizes the latent global carbon damage generated by complex electronic manufacturing lifecycles.

#7 about 3 min

Optimizing centralized cloud infrastructures for direct energy footprint reductions

Infrastructure cost-saving strategies completely parallel carbon reductions by actively shutting down abandoned instances and right-sizing underutilized virtual machine clusters.

#8 about 2 min

Interfacing open-source tooling for granular workload energy measurement

Integrating deep kernel diagnostic utilities enables dedicated telemetry metrics to trace power usage data down to singular operational applications.

#9 about 3 min

Tracking contextual software carbon intensities and service objectives

Application engineers can precisely calculate software carbon variance and outright halt automated build deployments if efficiency guidelines exceed tolerance thresholds.

#10 about 4 min

Carbon efficiency via automated schedule time and geographic shifting

Distributed systems naturally enable scheduling intensive batch jobs during optimal grid energy situations or directly moving workloads toward cleaner geographies.

#11 about 4 min

Analyzing cluster right-sizing and scaling experiments inside Kubernetes

Validating physical server node densities against scaled microservices discovers uniquely optimized configurations that drastically reduce wasted computational resources.

#12 about 2 min

Adapting carbon reduction systems for on-premises bare-metal configurations

Modifying automated public cloud methodologies to encompass localized infrastructure scaling and targeting internal architectural code bloat mitigates independent compute bottlenecks.

Matching moments

4:24 min

Balancing heavy compute demands with environmental sustainability goals

Mike Butcher Mike Butcher +3 · World Congress 2024

1:57 min

Connecting web performance optimization to ecological sustainability

Andreas Taranetz Andreas Taranetz · World Congress 2024

1:49 min

Coding for environmental sustainability in software engineering

Ulrich Irnich Ulrich Irnich · World Congress 2023

2:18 min

Evaluating artificial intelligence usage against structural climate impacts

Lorenzo Pieri · LIVE

2:09 min

Shifting focus from token maxing to AI cost management

Dona Sarkar +1 · Coffee With Developers

2:59 min

Improving Kubernetes cluster capacity planning and deployment carbon efficiency

Eric Deandrea Eric Deandrea · World Congress 2025